The SPE-GC/MSMS method with a LOQ of 0.15 µg/L was validated during our study by applying the requirements of French standard NF T 90-210 (NF 2018) using natural representative matrices (surface water, groundwater and tap water) under intermediate precision to demonstrate the reliability of the analytical data. The method was accredited by the French Accreditation Committee (COFRAC) in 2020.
The range of the calibration curve was studied with five calibration points ranging from 15 to 500 µg/L in a solvent mixture (v/v) of 80% DCM and 20% EtAc for 1,4-dioxane. The second order nonlinear internal standard calibration function was performed six times (on different days) from standard solutions freshly prepared each day. The correlation coefficients (r 2 ) obtained were ≥ 0.98. The back-calculated concentrations between the experimental and the nominal values must be within ± 15% for all calibration points, and within ± 20% for the calibration point corresponding to the LOQ at 0.15 µg/L. The results were acceptable and are presented in Fig. S2.
The limit of detection (LOD) was not considered in this study due to the low limit of quantification (LOQ) of 0.15 µg/L. The LOQ was defined as the lowest concentration of the analyte that can be determined with acceptable precision according to the French standard NF T 90-210 (NF 2018). The LOQ for 1,4-dioxane was validated under intermediate precision conditions in natural matrices. Six water samples (two groundwater, two surface water and two drinking water samples) were spiked at the pre-established LOQ. Inter-day precision was performed by analysing six series of duplicate extractions on six different days. To ensure the accuracy (trueness and precision), a maximum allowed tolerance (MAT) between the theoretical and the experimental values at the LOQ must not exceed ± 60%. The MAT was fixed as requested by the NF T90-210 standard, and its calculation was described in detail by Lardy-Fontan et al. (2018) (link) and Mirmont et al. (2023) (link). The LOQ of 0.15 µg/L for 1,4dioxane was validated.
Following the same procedure, the accuracy of the method was also evaluated for two intermediate concentrations of the calibration range (0.75 and 4 µg/L). In this case, the MAT did not exceed ± 40% for these two concentration levels.
The relative recovery study was carried out by spiking groundwater, surface water and drinking water at 0.15 µg/L, 0.75 µg/L and 4 µg/L in duplicates. Mean recoveries were calculated using the three matrices for each spiking concentration. As shown in Table 3S, the mean recoveries for 1,4-dioxane ranged from 117 to 114% for the three 1,4-dioxane concentrations studied. In addition, during the sampling campaign, several water samples of each batch were spiked with 1,4-dioxane at 1 µg/L in order to evaluate recoveries in the experimental conditions with different real matrix samples. Recovery results with their standard deviation are listed in Table 4S. It is worth noting that average recoveries are calculated with 92 different matrix samples (treated water, surface and groundwater) and over a period of 1 year and 4 months (reproducibility conditions).These experimental recoveries were within the limits (70% and 120%) set by ISO 21253-2:2019 (ISO 2019).
The relative uncertainty (U) was calculated in order to compare the measured results. The uncertainty was extended by a coverage factor (k) of 2 (95% confidence level). The measured uncertainty was 47% for the LOQ and 30% for the two intermediate concentrations in the calibration range.
The range of the calibration curve was studied with five calibration points ranging from 15 to 500 µg/L in a solvent mixture (v/v) of 80% DCM and 20% EtAc for 1,4-dioxane. The second order nonlinear internal standard calibration function was performed six times (on different days) from standard solutions freshly prepared each day. The correlation coefficients (r 2 ) obtained were ≥ 0.98. The back-calculated concentrations between the experimental and the nominal values must be within ± 15% for all calibration points, and within ± 20% for the calibration point corresponding to the LOQ at 0.15 µg/L. The results were acceptable and are presented in Fig. S2.
The limit of detection (LOD) was not considered in this study due to the low limit of quantification (LOQ) of 0.15 µg/L. The LOQ was defined as the lowest concentration of the analyte that can be determined with acceptable precision according to the French standard NF T 90-210 (NF 2018). The LOQ for 1,4-dioxane was validated under intermediate precision conditions in natural matrices. Six water samples (two groundwater, two surface water and two drinking water samples) were spiked at the pre-established LOQ. Inter-day precision was performed by analysing six series of duplicate extractions on six different days. To ensure the accuracy (trueness and precision), a maximum allowed tolerance (MAT) between the theoretical and the experimental values at the LOQ must not exceed ± 60%. The MAT was fixed as requested by the NF T90-210 standard, and its calculation was described in detail by Lardy-Fontan et al. (2018) (link) and Mirmont et al. (2023) (link). The LOQ of 0.15 µg/L for 1,4dioxane was validated.
Following the same procedure, the accuracy of the method was also evaluated for two intermediate concentrations of the calibration range (0.75 and 4 µg/L). In this case, the MAT did not exceed ± 40% for these two concentration levels.
The relative recovery study was carried out by spiking groundwater, surface water and drinking water at 0.15 µg/L, 0.75 µg/L and 4 µg/L in duplicates. Mean recoveries were calculated using the three matrices for each spiking concentration. As shown in Table 3S, the mean recoveries for 1,4-dioxane ranged from 117 to 114% for the three 1,4-dioxane concentrations studied. In addition, during the sampling campaign, several water samples of each batch were spiked with 1,4-dioxane at 1 µg/L in order to evaluate recoveries in the experimental conditions with different real matrix samples. Recovery results with their standard deviation are listed in Table 4S. It is worth noting that average recoveries are calculated with 92 different matrix samples (treated water, surface and groundwater) and over a period of 1 year and 4 months (reproducibility conditions).These experimental recoveries were within the limits (70% and 120%) set by ISO 21253-2:2019 (ISO 2019).
The relative uncertainty (U) was calculated in order to compare the measured results. The uncertainty was extended by a coverage factor (k) of 2 (95% confidence level). The measured uncertainty was 47% for the LOQ and 30% for the two intermediate concentrations in the calibration range.
